Fast characterization of non domestic load in urban wastewater networks by UV spectrophotometry

Abstract

Urban wastewater treatment plant efficiency, as well as biosolid quality, depends on urban wastewater quality, which can be affected by non domestic discharges (industrial, commercial etc.). The characterization of wastewater quality and non domestic discharge is complex, expensive and time consuming. However, these discharges must be controlled and reduced if possible. The development of a simple and fast methodology, namely based on alternative methods such as UV spectrophotometry, has been carried out and applied to different areas of a medium sized town of Southern Québec (Canada). Several autosamplers and on line/on site measurements have been used in critical control points of the network areas, for a dry weather campaign in four areas (industrial, commercial, hospital and university). The flow rate study, completed by the exploitation of conductivity measurements and the qualitative examination of UV spectra allows the discrimination of non domestic loads and their variability study from one point to another. The identification of critical discharges and organic shock loads has been possible with low investment, and mitigation actions have been proposed.

Enabling Automatic Detection of Anomalies in Wastewater: A Highly Simplified Approach to Defining “Normal” in Complex Chemical Mixtures

“…Yet "wastewater" is in reality a complex mixture of hundreds to thousands of different molecules, the concentrations of which can fluctuate based on community behaviors, industrial plant operations, precipitation (which can lead to dilution of wastewaters through stormwater infiltration), etc. (Thomas et al, 1999;Baurès et al, 2007;Tsoumanis et al, 2010;Schilperoort et al, 2012;Loos et al, 2013). Therefore, detecting an ability to properly detect anomalies is predicated on first being able to properly define the "normal" background against which to compare.…”

“…However, sudden changes can occur due to nonproblematic occurrences (e.g., operating hours of industrial or large office facilities). Therefore, a step toward defining "normal" has been to develop deterministic forecasting models based on external factors known to affect the chemical composition of wastewater, e.g.,time of day (Thomas, 2017), land use type (Lourenço et al, 2006;Baurès et al, 2007;Tsoumanis et al, 2010), and precipitation (Vaillant et al, 1999). Major limitations of such models are the complexity (i.e., relationship between factors and wastewater-for example, modeled behavior of citizens and businesses on a typical Monday must be differentiated from a holiday Monday, with numerous exceptional cases to be managed) and the need for data sources to ingest to drive the model.…”

Métodos Machine learning aplicados para estimar la concentración de los contaminantes de la DQO y de los SST en hidrosistemas de saneamiento urbano a partir de espectrometría UV-Visible

“…Figura 1-Comparación entre un sistema de alcantarillado combinado y separado (Brombach et al, 2005) ________________________________________________________________________ 29 Figura 2-Fuentes de los compuestos y sustancias presentes en aguas residuales ____________ 30 Figura 3-Clasificación de la DQO basada en la solubilidad y filtración (Field, 1987) ____________ 34 Figura 4-Tipos de muestreo y monitoreo (adaptado de González et al, 2009) ______________ 36 Figura 5-Atenuación de la radiación por una cubeta que contiene una solución (Thomas y Burgess, 2007) ________________________________________________________________________ 37 Figura 6-Relación entre caudal y espectros UV horarios obtenidos de aguas residuales de un hospital, comercio e industria de izquierda a derecha (Baurès et al, 2007) __________________ 39 Figura 7-Detección de diferentes parámetros de monitoreo en aguas a través del rango espectral UV-Visible (s::can Messtechnik GmbH, Viena, Austria). _________________________________ 42 Figura 8-Sonda disponibles en mercado para la medición in situ del espectro UV-Visible ______ 43 Figura 9-Esquema general de las simulaciones Monte Carlo _____________________________ 48 Figura 10-Pasos para generar una simulación Monte Carlo (Lepot, 2012) ___________________ 49 Figura 11-Detección de outliers en un conjunto de datos por medio de análisis bivariado ______ 50 Figura 12-Boxplot o diagrama de caja _______________________________________________ 51 Figura 13-PLS como un método de regresión lineal múltiple para la predicción de la propiedad y desde las variables X 1 , ..., X m , aplicando los coeficientes de regresión b 1 , ..., b m .…”